Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry

DES-13044; No of Pages 8 Desalination xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Desalination journal homepage: www.elsevier.com/locate/desal

Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry Minmin Zhang a,b, Qianhong She a,b, Xiaoli Yan a, Chuyang Y. Tang c,⁎ a b c

School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639798, Singapore Singapore Membrane Technology Centre, Nanyang Technological University, Singapore 639798, Singapore Department of Civil Engineering, The University of Hong Kong, Hong Kong

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Reverse solute diffusion (RSD) has important effect on FO scaling. • RSD of scaling precursors (Ca2 +, PO34 −, etc.) can promote severe Ca3(PO4)2 scaling. • RSD of anti-scaling agents (H+, EDTA, etc.) can suppress Ca3(PO4)2 scaling. • The effect of RSD of solute i is negligible if (Js/Jv)i ≪ (Cfeed)i.

a r t i c l e

i n f o

Article history: Received 19 April 2016 Received in revised form 3 August 2016 Accepted 12 August 2016 Available online xxxx Keywords: Forward osmosis (FO) Pressure retarded osmosis (PRO) Scaling control Reverse solute diffusion (RSD) Draw solution chemistry Anti-scaling precursors

a b s t r a c t We explored the specific role of reverse solute diffusion (RSD) on the scaling in osmotically-driven membrane processes (particularly forward osmosis (FO)). Both scaling precursors (e.g., Ca2+ and phosphate) and anti-scaling precursors (e.g., H+ and a chelating agent ethylenediamine tetraacetic acid (EDTA)) were used to investigate the effect of RSD and draw solution chemistry on calcium phosphate scaling. While draw solutions containing Ca2+ tend to promote calcium phosphate scaling, this effect was noticeable only if the specific RSD of Ca2+ (i.e., the ratio of Ca2+ flux to water flux) was greater than the original Ca2+ concentration in the feed water. The RSD of H+ and EDTA effectively suppressed scaling. For the first time, we demonstrated a new scaling control strategy for FO by the inclusion of anti-scaling functions in the draw solution chemistry. Our work has important implications for the design and operation of FO processes. © 2016 Elsevier B.V. All rights reserved.

1. Introduction ⁎ Corresponding author at: HW-619B, Department of Civil Engineering, The University of Hong Kong, Hong Kong. E-mail address: [email protected] (C.Y. Tang).

Osmotically driven membrane processes (ODMPs), including forward osmosis (FO) and pressure retarded osmosis (PRO), have gained more attention in the fields of seawater and brackish water desalination

http://dx.doi.org/10.1016/j.desal.2016.08.014 0011-9164/© 2016 Elsevier B.V. All rights reserved.

Please cite this article as: M. Zhang, et al., Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry, Desalination (2016), http://dx.doi.org/10.1016/j.desal.2016.08.014

2

M. Zhang et al. / Desalination xxx (2016) xxx–xxx

[1–3], wastewater reclamation [4,5], resource recovery [6,7], liquid food processing [8], and electricity generation [9,10]. Like reverse osmosis (RO) and nanofiltration (NF), FO may suffer from the scaling of sparingly soluble inorganic minerals [11], thus causing reduced water flux and shortened life span of membranes [12–14] or even the loss of membrane integrity [15,16]. Some common membrane scalants include calcium phosphate in wastewater reclamation and calcium carbonate and calcium sulphate in desalination [17,18]. According to the existing literatures, scaling of RO and NF membranes is affected by feed water chemistry (saturation index (SI), pH, and ionic strength), operating conditions (permeation flux, applied pressure, crossflow velocity, and temperature), and the presence of other competing agents such as antiscalants and organics with chelating properties [14,19–21]. Similar behaviors have been reported for FO scaling [22–26]. However, scaling in ODMPs can be further complicated by internal concentration polarization (ICP) and reverse solute diffusion (RSD). Zhang et al. [27] reported that PRO scaling in active-layer facing draw solution orientation (AL-DS) was greatly enhanced by (1) the RSD of scaling precursors from the DS to the feed solution (FS) and (2) the concentrative ICP of both draw solutes (through RSD) and feed solutes inside the porous membrane substrate. Although concentrative ICP is not important in the active-layer facing feed solution orientation (ALFS, which is commonly used for FO applications), this orientation will still suffer from the RSD effect [11]. Both inorganic and organic DSs have been reported in the literature [28,29]. According to some recent reviews [29–31], the major requirements for an ideal DS include high osmotic pressure, high water solubility, low viscosity, high diffusivity, and easy regeneration. Nevertheless, its propensity in promoting membrane fouling should not be overlooked [27,32,33]. It has been reported that some specific ions (e.g., SO24 − and Ca2 +) in DS can act as precursors to promote severe membrane scaling during PRO operation [27]. However, the exact conditions under which this effect is important has not been systematically investigated. In addition, the water chemistry of a DS can also potentially affect the pH of the feed solution (e.g., via the RSD of H + or OH−), which can either reduce or promote scaling. Conceptually, by carefully tailoring DS chemistry and imparting anti-scaling/antifouling functionalities to the DS, one may be able to control scaling (as well as other types of fouling) in ODMPs effectively – an option that is never available for pressure-driven membrane processes. Such considerations may open a new dimension for DS design and selection. To the authors' best knowledge, the use of DS chemistry for reducing FO scaling has not been reported previously. The objective of this paper is to systematically investigate the specific role of RSD on FO scaling. Calcium phosphate, a commonly found sparingly soluble mineral in wastewater, was used as model scalant due to its significance in membrane-based wastewater reclamation [24,34,35]. The role of specific ions in DS and the inclusion of anti-scaling chemistry were systematically tested. The results of the current study may provide important insights into FO scaling and scaling control. 2. Material and methods

Table 1 Compositions and properties of feed solutions. Feed solution

NaCl (mM)

CaCl2 (mM)

Na3PO4 (mM)

pH Osmotic pressure (bar)a

Baseline tests Scaling tests w/o Ca2+ in FS Scaling tests with Ca2+ in FS

61 35

0 0

0 20

7.5 2.8 7.5 2.8

30

12

0.8

7.5 2.8

a All the feed solutions have identical osmotic pressure. The osmotic pressure is calculated using the OLI's Stream Analyzer 3.1 software (OLI Systems, Inc., Morris Plains, NJ).

Stream Analyzer 3.1 (OLI System, Inc., Morris Plains, NJ) was used for the calculation of osmotic pressure values. 2.2. Membranes Two commercial FO membranes (a cellulose triacetate membrane denoted as CTA and a thin film composite polyamide membrane denoted as TFC) used in the study were obtained from Hydration Technologies Inc. (HTI, Albany, OR). The TFC membrane had a higher water permeability and solute rejection (Table 3) but also a greater surface roughness [36–38]. In comparison, the CTA membrane had a much smoother surface [37]. It had been well documented in the literature that a rougher membrane surface tends to promote fouling [11,39]. 2.3. Bench-scale FO scaling tests A crossflow FO setup was modified from our existing study [40]. Briefly, an FO membrane coupon with an effective area of 42 cm2 was placed in a CF042 (Sterlitech) cell. Diamond-patterned net-type spacers were inserted into both the FS and DS channels. A crossflow velocity of ~9.7 cm/s was used for both the FS and DS. The temperature of both feed water and draw solution were at ambient temperature (25 ± 1 °C). The weight change of DS was measured by a digital balance, which was used for FO water flux calculation. All scaling tests were performed in the AL-FS orientation. A new piece of membrane was used for each scaling test. After the 20-min conditioning of the membrane coupon by a calcium-free feed solution in test cell, a desired amount of CaCl2 was introduced into the FS to initiate the scaling test. Where needed, some minor pH adjustment was also performed immediately after the dosage of CaCl2 to ensure the solution pH was maintained at the desired value. The scaling test was then continued for another 8 h and membrane flux was measured by monitoring the weight change of the feed solution using a digital balance. Baseline tests were also performed using feed solutions of identical osmotic pressure (using NaCl solution) and pH (Table 1). The deviation of water flux between the scaling tests from the corresponding scaling-free baseline level can be attributed to FO scaling [41]. To study the effect of chemistry of different draw solutions, their compositions were chosen in such a way to ensure an identical initial membrane flux (see detailed DS composition in Table 2). Scaled membranes were further characterized using a Joel JSM 7600F Field Emission Scanning Electron Microscope (SEM). Air-dried membrane coupons were sputter-coated with a thin

2.1. Chemicals and solutions Unless specified otherwise, all solutions were prepared with ACS grade chemicals and MilliQ water. FS and DS were prepared by the addition of pre-determined amounts of NaCl, CaCl2, Na3PO4, NH4H2PO4, (NH4)2HPO4, and Ethylenediaminetetraacetic acid disodium salt (EDTA). Their solution pHs were adjusted using NaOH and HCl. In the current study, the EDTA was chosen as a model chelating agent to sequestrate calcium from participating in scaling formation. The DS compositions were chosen in such a way as to achieve an identical initial FO water flux (17.5 ± 0.8 L/m2·h). Detailed compositions and properties of the FSs and DSs are summarized in Tables 1 and 2, respectively. OLI's

Table 2 Compositions and properties of draw solutions. Draw solution

Concentration

pH

Osmotic pressure (bar)a

NaCl CaCl2 NH4H2PO4 + NaCl (NH4)2HPO4 + NaCl EDTA + NaCl

2.0 M 1.3 M 1.0 M + 1.0 M 1.0 M + 1.0 M 0.1% + 1.0 M

4.0, 7.0, or 8.5 7.0 4.5 8.5 7.0

100 116 92 94 100

a The draw solutions were chosen in such a way that an identical initial FO flux of 17.5 ± 0.8 L/m2·h was obtained for all tests. The osmotic pressure is calculated using the OLI's Stream Analyzer 3.1 software (OLI Systems, Inc., Morris Plains, NJ).

Please cite this article as: M. Zhang, et al., Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry, Desalination (2016), http://dx.doi.org/10.1016/j.desal.2016.08.014

M. Zhang et al. / Desalination xxx (2016) xxx–xxx

3

Table 3 Separation properties of CTA and TFC FO membranes.a Membrane

ANa (m/s·Pa)a

BNa (m/s)a

ACa (m/s·Pa)a

BCa (m/s)a

NaCl rejection (%)a

CaCl2 rejection(%)a

Js, Ca (mol/m2·h)b

CTA TFC

2.0E-12 6.3E-12

1.2E-7 1.8E-7

2.0E-12 6.0E-12

1.2E-7 1.5E-7

93.5 94.2

94.9 97.0

0.055 0.019

a b

The CTA and TFC membrane properties (salt rejection, A and B value) were measured using a reverse osmosis system with a feed solution containing 10 mM CaCl2 or 10 mM NaCl. The Ca2+ reverse solute flux of CTA and TFC membranes were tested in a FO system where pure MQ water was used as feed solution and 1.3 M CaCl2 was used as draw solution.

layer (~5 nm) of platinum and were examined at an acceleration voltage of 5 kV. 3. Results and discussion 3.1. The effect of RSD of scaling precursors As a unique feature for ODMPs, FO requires a high concentration DS to drive water flow during which some draw solutes can diffuse though the membrane into the FS. In the current section, the effect of Ca2+ as a scaling precursor was evaluated using the CTA membrane in AL-FS orientation. In Fig. 1a, membrane scaling was performed using a feed solution containing no calcium but with phosphate (i.e., 35 mM NaCl and

20 18

Water flux (L/m2-h)

16 14 12 10 8 6 4

FS: Solution a (0 mM Ca) DS Baseline Scaling CaCl2

2

(a)

0

0

60

120

180

240

300

360

420

480

360

420

480

Time (min)

20 18

Water flux (L/m2-h)

16 14 12 10 8 6 FS: Solution b (12 mM Ca) DS Baseline Scaling CaCl2

4 2

20 mM Na3PO4, see Table 1). The DS contained 1.3 M CaCl2. Phosphate in the feed solution could precipitate with calcium under RSD from the DS. In parallel, a baseline test was performed using a feed solution that was free of both calcium and phosphate to eliminate membrane scaling (Table 1). Compared to the baseline test, there was approximately 20% flux decline for the scaling test, which can be attributed to the RSD of Ca2+ [41,42]. The diffusion of Ca2+ can effectively increase the Ca2+ concentration near the FS-membrane interface, which leads to accelerated FO scaling due to the co-presence of calcium (through RSD) with phosphate (from the original feed solution). Fig. 1b presents the case where the feed contains a 12 mM CaCl2 (along with other background salts such as NaCl and Na3PO4 to obtain the target feed osmotic pressure of 2.8 bar, see Table 1). In this case, even when the DS only contained NaCl, significant membrane scaling occurred (58 ± 6% flux drop compared to its baseline, also see SEM micrographs in Appendix A), since the FS contained abundant amount of both calcium and phosphate. When CaCl2 was used as the DS, a similar relative flux decline was observed (48 ± 6%, Appendix B). In this case, the presence of scaling precursor Ca2+ had no apparent effect in promoting scaling. Thus, the results in Fig. 1b appears to contradict with those in Fig. 1a as well as some other earlier studies [27,32,41] revealing the critical role of divalent ions (Ca2+ and Mg2+) in DS on mediating membrane fouling and scaling. This apparent contradiction can be reconciled by considering the relative dominance of RSD of Ca2+ into FS over the amount of Ca2+ that was originally present in the FS. According to prior studies [38,40,43], the effective draw solutes concentration arising from RSD can be represented by the ratio of reverse solute flux Js over FO water flux Jv. In the current study, Js/Jv for calcium was estimated to be ~ 3.1 mM for the CTA membrane (Table 3). When there is little amount of Ca2+ in the original FS (e.g., 0 mM in Fig. 1a), the scaling behavior is dominated by Ca2+ diffused from the DS into the FS. In contrast, when there is relatively high concentration of Ca2 + in the original FS (e.g., 12 mM in Fig. 1b), the additional contribution from RSD becomes less important, leading to an indiscernible effect of Ca2+ in the DS. To further understand the relative importance of RSD of Ca2+ versus Ca2+ contained in the original feed solution, a mass transport model was developed in Appendix C. According to this model, the concentration of a generic solute i (such as a scaling precursor Ca2+) near the FS-membrane interface is given by fECP ∙ (Cfeed)i + (fECP − 1)(Js/ Jv)i, where fECP is the external concentration polarization factor, and the terms (Cfeed)i and (Js/Jv)i represent the contribution from the original FS and RDS, respectively. Clearly, the RSD has negligible effects to the FS chemistry if (Js/Jv)i ≪ (Cfeed)i. The result in the current study is consistent with Xiao et al. [43] who suggested that the RSD contribution to overall salt accumulation in an FO bioreactor is negligible when Js/Jv is much less than the feed solute concentration Cfeed.

NaCl

0

(b)

0

60

120

180

240

300

Time (min) Fig. 1. The effect of RSD of Ca2+ on calcium phosphate scaling. a) FS free of Ca2+ (35 mM NaCl and 20 mM Na3PO4); b) FS containing Ca2+ (30 mM NaCl, 12 mM CaCl2 and 0.8 mM Na3PO4). Other experimental conditions: FS pH 7.5 ± 0.1, DS pH 7.0 ± 0.1, AL-FS orientation, CTA membrane, crossflow velocity ~9.7 cm/s, and temperature 25 ± 1 °C. The error bars were based on two independent experiments.

3.2. The effect of RSD of anti-scaling agents Whereas the RSD of Ca2+ tends to promote scaling, we also explore the beneficial effect of RSD on scaling control in ODMPs. For example, a lower pH in the FS is known to alleviate scaling (Appendix D). Accordingly, one can take advantage of the RSD of specific chemicals such as H+ from the DS into FS in order to control the FS solution pH and thus to reduce scaling. Fig. 2 presents the FO water flux behavior using a

Please cite this article as: M. Zhang, et al., Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry, Desalination (2016), http://dx.doi.org/10.1016/j.desal.2016.08.014

4

M. Zhang et al. / Desalination xxx (2016) xxx–xxx

20 18

Water Flux (L/m2-h)

16 14 12 10 8 6 Baseline pH 4 pH 7 pH 8.5

4 2 0

0

60

120

180

240

300

360

420

480

Time (min) Fig. 2. The effect of DS pH on calcium phosphate scaling. Other experimental conditions: FS containing 12 mM Ca2+ and 0.8 mM PO3− at pH 7.5 ± 0.1; DS: 2 M NaCl; AL-FS 4 orientation; CTA membrane; crossflow velocity ~9.7 cm/s; and temperature 25 ± 1 °C. The error bars were based on two independent experiments.

20

20

18

18

16

16

14

14

Water flux (L/m2-h)

Water flux (L/m2-h)

2 M NaCl DS adjusted to different pH values (pH 4, 7, and 8.5). Severe scaling occurred when the pH of the DS was at 7 or 8.5. When the DS pH was reduced to 4, the scaling flux curve overlapped with the baseline, which suggests that membrane scaling was negligible under this condition (also see Appendix A). The improved anti-scaling performance and flux stability can be attributed to the RSD of H+. The significantly higher H+ concentration in the DS at pH 4 allows their faster diffusion through the membrane to modify the local FS pH condition, and the resultant pH reduction near the membrane surface effectively prevent calcium phosphate scaling by shifting the phosphate specifica− tion (HPO2− 4 → H2PO4 , see Appendix D). It is further interesting to note that, despite the use of different pH for the DS (pH 4, 7, and 8.5), the bulk FS solution pH for three cases was nearly identical (pH 7.5). Thus, instead of relying on the modification of the bulk FS pH, the RSD of draw solutes can cause localized pH change at the membrane surface to affect scaling behavior. In addition, the RSD of H+ is accompanied with co-ions (Cl− in the current study) to maintain the solution neutrality. Future studies shall further investigate the potential role of RSD coions in the context of ODMPs. The effect of EDTA as an anti-scaling agent was also investigated by the addition of 0.1% EDTA to the DS (Fig. 3). Clearly, the inclusion of

EDTA in the DS significantly improved membrane anti-scaling performance, which can be attributed to the sequestration of Ca2+ by EDTA. It is well known in the literature that the addition of EDTA to FS can effectively reduce calcium-based scaling by forming the EDTA-Ca2+ complex, making Ca2 + less available for scaling formation in the feed solution [44–46]. In the current study, EDTA was introduced in the DS. Nevertheless, its RSD into the FS can still effectively alleviate membrane scaling, which opens a new dimension for scaling control in ODMPs. Compared to the use of acidic draw solution of pH 4, the effect of EDTA on scaling control was mild for the RO-like FO membrane in the current study. EDTA is relatively bulky and contains multiple charges, which results in a low RSD (Js/Jv ~ 0.1 g/L, an order of magnitude lower compared to NaCl [47]). Despite that, the use of EDTA for FO scaling control can be promising in view of the expanding interests over lower-rejection (such as NF-like [48–52] and UF-like [53,54]) FO membranes. Although the current study focused on EDTA as a model ligand, the underlining mechanism of ligand chemistry can be applied to other ligands such as citric acid and oxalic acid. Future systematic studies are needed to further optimize the matching of draw solutes and FO membranes, with due consideration to fouling behavior in addition to water flux and rejection performance. In certain cases, the DS may simultaneously contain scaling precursor as well as anti-scaling agents. We explored the use of NH4H2PO4 and (NH4)2HPO4 as DSs in the current study (Table 2). Both NH4H2PO4 and (NH4)2HPO4 contains phosphate as a scaling precursor to calcium phosphate scaling. In addition, the NH4H2PO4 solution is relatively acidic (pH 4.5), which provides a potential source for H+ for scaling control. Fig. 4 compares the FO water flux behavior using DSs with scaling precursors (NH4H2PO4 and (NH4)2HPO4) to that without scaling precursors (NaCl). The use of NaCl as DS resulted in approximately 50% flux decline compared to the baseline over the 8-hour test. In contrast, scaling was much more severe when (NH4)2HPO4 was used as the DS: FO flux was nearly completely lost in b4 h. As discussed in Section 3.1, the RSD of a scaling precursor (in this case, phosphate) can significantly increase its concentration in the FS and thus accelerate membrane scaling. Interestingly, the use of NH4H2PO4 as the DS led to a very stable FO water flux. In the latter case, despite that the DS contained a very high phosphate concentration, scaling was not severe. This phenomenon can be attributed to the role of RSD in modifying the local pH in the FS. The relatively acidic NH4H2PO4 solution (pH 4.5) contains significantly amount of H+, whose diffusion into the FS helps to mitigate calcium phosphate scaling. In the current study, the pH effect seems to dominate over the phosphate precursor effect for the case of NH4H2PO4, possibly due to

12 10 8 6 4

Baseline NaCl NaCl+0.1% EDTA

2 0

0

60

120

12 10 8 6

Baseline NH4H2PO4(pH 4.5)

4

NaCl (pH 8.5) (NH4)2HPO4 (pH 8.5)

2 180

240

300

360

420

480

Time (min) Fig. 3. The effect of the addition of EDTA in DS on calcium phosphate scaling. Other experimental conditions: FS containing 12 mM Ca2+ and 0.8 mM PO3− at pH 7.5 ± 0.1; 4 DS at pH 7.0 ± 0.1; AL-FS orientation; CTA membrane; crossflow velocity ~9.7 cm/s; and temperature 25 ± 1 °C. The error bars were based on two independent experiments.

0

0

60

120

180

240

300

360

420

480

Time (min) Fig. 4. The effect of phosphate based draw solution on calcium phosphate scaling. Other experimental conditions: FS containing 12 mM Ca2+ and 0.8 mM PO3− at pH 7.5 ± 0.1; 4 DS at pH 7.0 ± 0.1; AL-FS orientation; CTA membrane; crossflow velocity ~9.7 cm/s; and temperature 25 ± 1 °C. The error bars were based on two independent experiments.

Please cite this article as: M. Zhang, et al., Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry, Desalination (2016), http://dx.doi.org/10.1016/j.desal.2016.08.014

M. Zhang et al. / Desalination xxx (2016) xxx–xxx

the small size of H+ and thus its faster diffusion through the FO membrane. In contrast, the pH of (NH4)2HPO4 was approximately 8.5, limiting the beneficial effect of H+ RSD.

3.3. The effect of membrane properties Fig. 5 compares the scaling behavior between the CTA and TFC membranes. In Fig. 5a, NaCl was used as the DS, and the FS contained 12 mM calcium. Although both membrane approached to a similar final flux after the 8-hour scaling experiment, the initial flux decline was more rapid for the TFC membrane. The initial difference may be attributed to the surface properties of the TFC membrane: (1) The polyamidebased TFC membrane surface is rich in carboxylic groups (–COO−), which can complex with Ca2 + and serve as sites for initiating membrane scaling [55]; (2) the TFC membrane has greater surface roughness that helps to accelerate fouling and scaling [56]. A similar observation was reported by Gu et al. [56] that CTA membranes had better antifouling performance against organic fouling in FO. The similar flux behavior of the two membranes towards the end of scaling tests can be explained

20 18

Water flux (L/m2-h)

16 14 12 10 8 FS: Solution b (12 mM Ca) DS: NaCl Membrane Baseline Scaling TFC CTA

6 4 2 0

0

60

120

180

240

300

360

420

480

360

420

480

Time (min)

(a) 20 18

5

by the dominance of foulant-foulant interaction over foulant-membrane interaction [40]. Fig. 5b presents the scaling behavior for the two membranes using a DS containing scaling precursors (i.e., CaCl2) and a Ca2+ free FS. In this case, we observed an opposite trend to that in Fig. 5a. Despite the more favorable surface properties of the CTA membrane, it suffered more flux decline (~20%) in the current case. In comparison, the flux decline for the TFC membrane was much less obvious. This peculiar result can be explained by the dominance of RSD in this particular case: since the FS contained no calcium originally, the membrane scaling was critically controlled by RSD of Ca2+. Compared to the TFC membrane, the CTA membrane had lower solute rejection (Table 3) and thus it had a greater propensity of RSD of Ca2+. In the current study, the (Js/Jv)Ca2+ for the CTA membrane was approximately 3.1 mM, which was much larger than that for the TFC membrane (~1.1 mM). 4. Implications The current study systematically revealed the critical importance of RSD on membrane scaling and fouling in ODMPs. RSD is unique for ODMPs, and it represents a pathway for the draw solutes to influence the FS chemistry, which may either promote or abate fouling/scaling. As illustrated in the conceptual diagram in Fig. 6, the effect of RSD on membrane scaling depends on the feed solution chemistry, draw solution chemistry, and the membrane rejection properties. In general, scaling precursors contained in the DS have the tendency to promote severe scaling. The current study further reveals that the relative importance of scaling precursors is governed by (Js/Jv)i of the particular precursor i in comparison to its feed solution concentration (Cfeed)i – the scaling promoting effect is less prominent when its specific RSD (Js/Jv)i ≪ (Cfeed)i. Therefore, from scaling control point of view, it is generally a good practice to avoid DSs that contain significant concentrations of scaling pre3− cursors such as Ca2 +, SO24 −, CO2– 3 and PO4 . Where such precursors are unavoidable, one should consider limiting their specific RSD values to an acceptable value. Considering the results from the current study as well as early literature [43], (Js/Jv)i ≪ 0.1(Cfeed)i may be regarded as a rule of thumb. In this case, a membrane with sufficiently high rejection should be selected to limit the undesirable effect of RSD. Membrane properties play critical role in the scaling and fouling behavior. The current study confirms the greater scaling tendency of the TFC membrane, probably as a result of its rougher membrane surface and the presence of abundant –COO− groups. Where the DS contains scaling precursors, their rejection by the membrane is a very important consideration. In this regard, membrane surface coating and modification may simultaneous improve the antifouling properties and

Water flux (L/m2-h)

16 14 12 10 8 6

FS: Solution a (0 mM Ca) DS: CaCl2

4

Memrbane Baseline Scaling TFC CTA

2 0

0

60

120

180

240

300

Time (min)

(b) Fig. 5. The effect of membrane properties on calcium phosphate scaling. a) FS containing 12 mM Ca2+ and DS containing 2 M NaCl; b) FS containing 0 mM Ca2+ and DS containing 1.3 M CaCl2. Other experimental conditions: AL-FS orientation; crossflow velocity ~ 9.7 cm/s; and temperature 25 ± 1 °C. The error bars were based on two independent experiments.

Fig. 6. Conceptual diagram illustrating the role of RSD in promoting or reducing scaling in ODMPs.

Please cite this article as: M. Zhang, et al., Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry, Desalination (2016), http://dx.doi.org/10.1016/j.desal.2016.08.014

M. Zhang et al. / Desalination xxx (2016) xxx–xxx

improving the membrane rejection properties [57], although a right balance between membrane permeability and rejection need to be carefully controlled [38]. The current study, for the first time, investigated the exploitation of DS chemistry for scaling control. Using calcium phosphate as a model scalant, we demonstrated that the RSD of anti-scaling agents, such as [H+] and EDTA, was highly effective in suppressing scaling in ODMPs. Unlike conventional pressure driven membrane processes, RSD is unavoidable in ODMPs. Although RSD has been traditionally considered undesirable in FO [32,33], a well designed DS chemistry can indeed play an active role in scaling abatement (Fig. 6). Furthermore, one can potentially rely on the DS chemistry to selectively modify the local FS solution chemistry near the membrane surface without the need to change the bulk FS chemistry, which can potentially translate into significant reduction in chemical dosage. The ability to deliver the chemicals (e.g., H+ and EDTA) to where they are needed opens an unprecedented new dimension for scale control in ODMPs. 5. Conclusions The current study investigated the specific role of RSD on FO scaling by calcium phosphate. The presence of Ca2+ in the DS tends to promote scaling. In contrast, draw solutions with anti-scaling chemistry (e.g., low pH or the inclusion of EDTA) can potentially reduce FO scaling as a result of their RSD into the FS. The effect RSD was only noticeable when its specific RSD (Js/Jv)i dominated over the corresponding feed concentration. In addition to the surface properties of membranes, their rejection properties can also significantly affect the FO scaling behavior through their role on controlling the specific RSD of scaling precursors (or anti-precursors). Understanding of these competing mechanisms is critical for the development of improved DS chemistries, membrane materials, and process operation strategies for scaling and fouling control in ODMPs. Acknowledgements The authors thank the Singapore Ministry of Education (Grant #MOE2011-T2-2-035, ARC 3/12) and the Strategic Research Theme (Clean Energy) of the University of Hong Kong for the partial financial support of the work. The authors are also grateful to HTI for supplying membrane samples. Appendix A. SEM results Fig. 7 shows the SEM micrographs of membranes after scaling tests. The baseline case was tested using a feed solution of 61 mM NaCl at pH 7.5 together with a DS of 2 M NaCl at pH 7. As expected, the membrane at the baseline conditions was free of scaling (Fig. 7(a)). In Fig. 7(b), the feed solution contained 30 mM NaCl, 12 mM CaCl2, and 0.8 mM Na3PO4 at pH 7.5 and the DS was 2 M NaCl at pH 7. The inclusion

(a) Baseline test

1.0

0.8

Normalized flux (-)

6

0.6

0.4

FS: Solution b (12 mM Ca) DS Scaling CaCl2

0.2

NaCl

0.0

0

60

120

180

240

300

360

420

480

Time (min) Fig. 8. The effect of Ca2+ RSD of DS on calcium phosphate scaling. Other scaling experimental conditions (identical to the conditions for Fig. 1b): FS: pH 7.5 ± 0.1, 12 mM Ca2+ and 0.8 mM PO3− 4 ; DS: pH 7.0 ± 0.1, AL-FS orientation, CTA membrane applied, crossflow velocity ~ 9.7 cm/s, and temperature 25 ± 1 °C. The error bars represent the standard deviations based on two independent experiments.

of both calcium and phosphate in the feed solution resulted in severe scaling formation. However, when the pH of the draw solution was reduced to pH 4.0, membrane scaling was nearly eliminated (with only minor patches of scalant deposition, see Fig. 7(c)). The current study clearly demonstrates that the RSD of H+ was effectively in controlling calcium phosphate scaling. Appendix B. Effect of Ca2+ in the DS on the normalized flux behavior Fig. 8 presents the normalized flux where the feed contains a 12 mM CaCl2. The results show that there is a 48 ± 6% flux drop for NaCl DS and a 58 ± 6% flux drop for CaCl2 DS respectively. Appendix C. Modeling the role of RSD on membrane scaling A mass transfer model is included in this appendix to determine the relative importance of RSD in comparison to the feed solution concentration of a generic solute i for the AL-FS membrane orientation (Fig. 9). Readers may also refer to reference [27] for the model of generic solute i in the AL-DS orientation in the context of PRO scaling/fouling and reference [43] for the description of the overall solute concentration in the context of FO membrane bioreactor. Fig. 9 presents a mass balance for the solute of interest i in the control volume (represented by the dotted box) near the vicinity of the membrane surface. For the simplicity of presentation, we will drop the subscript i in the model development, but understanding that the equations are applicable to an individual solute species i as well as the sum of

(b) Scaling test (DS at pH 7)

(c) Scaling test (DS at pH 4)

Fig. 7. SEM micrographs of membranes after scaling tests. The baseline test (a) was performed using a feed solution of 61 mM NaCl at pH 7.5 and a DS of 2 M NaCl at pH 7. The scaling tests were performed with a feed solution containing 30 mM NaCl, 12 mM CaCl2, and 0.8 mM Na3PO4 at pH 7.5, using a DS of 2 M NaCl at either pH 7 (b) or pH 4 (c).

Please cite this article as: M. Zhang, et al., Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry, Desalination (2016), http://dx.doi.org/10.1016/j.desal.2016.08.014

M. Zhang et al. / Desalination xxx (2016) xxx–xxx

7

1.0 0.9 0.8 0.7 H3PO4

0.6

-

H2PO4

0.5

2-

HPO4

0.4

3-

PO4

0.3 0.2 0.1 0.0

0

1

2

3

4

5

6

7

8

9

10 11 12 13 14

pH Fig. 10. The hydrolysis of phosphate versus pH.

Fig. 9. Modeling the relative contribution of reverse solute flux to the feed solution chemistry in the vicinity of the FS-membrane interface.

these individual species. Solutes can enter the control volume by either convection from the feed solution (JvC) or through RSD from the draw solution (Js). The accumulated solutes in the vicinity of the membrane surface has to be balanced by the diffusion of the solutes back to the bulk feed solution (D·dC/dx), a phenomenon commonly known as the ECP effect in the context of ODMPs. Therefore,

D

dC ¼ Jv C þ Js dx

ð1Þ

where the distance x is measured from the hydrodynamic boundary layer whose thickness is δ. Thus, the following boundary conditions are applicable: C ¼ C feed at x ¼ 0

ð2Þ

and C ¼ C m at x ¼ δ

where fECP is the external concentration polarization factor:   J δ f ECP ¼ exp v D

ð7Þ

It is worthwhile to note that Eq. (6) takes a similar form to the model developed by Zhang et al. [27] in the context of PRO as well as the model developed by Xiao et al. [43] in the context of salt accumulation in FO membrane bioreactors. In all the cases, the concentration at the membrane surface is the superimposed contributions of the feed solutes and the solutes from RSD. Indeed, Eq. 6 is also consistent with the ECP model for RO [58] noting that a forward solute flux can be mathematically treated as a negative reverse solute flux. Eq. (6) implies that the effect of RSD is not important if Js/Jv ≪ Cfeed. Appendix D. The effect of feed solution pH Phosphoric acid has four hydrolysis bases in water: non-charged phosphoric acid, dihydrogen phosphate ion, hydrogen phosphate and phosphate. The detailed distribution depending on the water pH is presented in Fig. 10. For pH above 8, the dominating ion is

ð3Þ

20

Eq. (1) can be rearranged as D

dðC þ J s =J v Þ ¼ J v ðC þ J s =J v Þ dx

ð4aÞ

or dðC þ J s =J v Þ J v ¼ dx ðC þ J s =J v Þ D

ð4bÞ

By integrating Equitation (4b) from x = 0 to x = δ and substituting the boundary conditions, we have   C m þ J s =J v J δ ¼ exp v D C feed þ J s =J v

12

8

4

DS: NaCl Baseline: pH 4 pH 7.5

0 0

ð5Þ

or C m ¼ f ECP C feed þ ð f ECP −1Þ  J s =J v

Water Flux (L/m2-h)

16

ð6Þ

60

120

180

240

300

360

420

480

Time (min) Fig. 11. The effect of FS pH on calcium phosphate scaling. Other scaling experimental conditions: FS: Solution b (12 mM Ca2+ and 0.8 mM PO3− 4 ); DS: 2 M NaCl, pH 7.0 ± 1; AL-FS orientation, CTA membrane applied, crossflow velocity ~ 9.7 cm/s, and temperature 25 ± 1 °C. The error bars represent the standard deviations based on two independent experiments.

Please cite this article as: M. Zhang, et al., Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry, Desalination (2016), http://dx.doi.org/10.1016/j.desal.2016.08.014

8

M. Zhang et al. / Desalination xxx (2016) xxx–xxx

HPO24 −. The H2PO− 4 becomes dominating for pH around 3 to 6. At approximately neutral pH level, both HPO24 − and H2PO− 4 coexist in the solution. The effect of FS pH (4 and 7.5) on calcium phosphate scaling is shown in Fig. 11. The flux behavior for a FS pH 7.5 demonstrated a nearly 50% flux deduction. Membrane scaling was not discernible at pH 4. References [1] R.E. Kravath, J.A. Davis, Desalination of sea water by direct osmosis, Desalination 16 (1975) 151–155. [2] S.F. Zhao, L. Zou, D. Mulcahy, Brackish water desalination by a hybrid forward osmosis–nanofiltration system using divalent draw solute, Desalination 284 (2012) 175–181. [3] S.C. Low, Preliminary studies of seawater desalination using forward osmosis, Desalin. Water Treat. 7 (2012) 41–46. [4] R. Valladares Linares, Z. Li, S. Sarp, S.S. Bucs, G. Amy, J.S. Vrouwenvelder, Forward osmosis niches in seawater desalination and wastewater reuse, Water Res. 66 (2014) 122–139. [5] C. Boo, M. Elimelech, S. Hong, Fouling control in a forward osmosis process integrating seawater desalination and wastewater reclamation, J. Membr. Sci. 444 (2013) 148–156. [6] J. Zhang, Q. She, V.W.C. Chang, C.Y. Tang, R.D. Webster, Mining nutrients (N, K, P) from urban source-separated urine by forward osmosis dewatering, Environ. Sci. Technol. 48 (2014) 3386–3394. [7] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Toward resource recovery from wastewater: extraction of phosphorus from digested sludge using a hybrid forward osmosis–membrane distillation process, Environ. Sci. Technol. Lett. 1 (2014) 191–195. [8] K.B. Petrotos, P. Quantick, H. Petropakis, A study of the direct osmotic concentration of tomato juice in tubular membrane - module configuration. I. The effect of certain basic process parameters on the process performance, J. Membr. Sci. 150 (1998) 99–110. [9] K.L. Lee, R.W. Baker, H.K. Lonsdale, Membranes for power generation by pressureretarded osmosis, J. Membr. Sci. 8 (1981) 141–171. [10] A. Achilli, T.Y. Cath, A.E. Childress, Power generation with pressure retarded osmosis: an experimental and theoretical investigation, J. Membr. Sci. 343 (2009) 42–52. [11] Q. She, R. Wang, A.G. Fane, C.Y. Tang, Membrane fouling in osmotically driven membrane processes: a review, J. Membr. Sci. 499 (2016) 201–233. [12] W.-Y. Shih, A. Rahardianto, R.-W. Lee, Y. Cohen, Morphometric characterization of calcium sulfate dihydrate (gypsum) scale on reverse osmosis membranes, J. Membr. Sci. 252 (2005) 253–263. [13] C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, State-of-the-art of reverse osmosis desalination, Desalination 216 (2007) 1–76. [14] A. Antony, J.H. Low, S. Gray, A.E. Childress, P. Le-Clech, G. Leslie, Scale formation and control in high pressure membrane water treatment systems: a review, J. Membr. Sci. 383 (2011) 1–16. [15] Y.N. Wang, E. Järvelä, J. Wei, M. Zhang, H. Kyllönen, R. Wang, C.Y. Tang, Gypsum scaling and membrane integrity of osmotically driven membranes: the effect of membrane materials and operating conditions, Desalination 377 (2016) 1–10. [16] M. Xie, C.Y. Tang, S.R. Gray, Spacer-induced forward osmosis membrane integrity loss during gypsum scaling, Desalination (2016) (in press). [17] G. Greenberg, D. Hasson, R. Semiat, Limits of RO recovery imposed by calcium phosphate precipitation, Desalination 183 (2005) 273–288. [18] F. He, K.K. Sirkar, J. Gilron, Studies on scaling of membranes in desalination by direct contact membrane distillation: CaCO3 and mixed CaCO3/CaSO4 systems, Chem. Eng. Sci. 64 (2009) 1844–1859. [19] A. Al-Amoudi, R.W. Lovitt, Fouling strategies and the cleaning system of NF membranes and factors affecting cleaning efficiency, J. Membr. Sci. 303 (2007) 4–28. [20] A.S. Al-Amoudi, Factors affecting natural organic matter (NOM) and scaling fouling in NF membranes: a review, Desalination 259 (2010) 1–10. [21] T. Waly, M.D. Kennedy, G.J. Witkamp, G. Amy, J.C. Schippers, The role of inorganic ions in the calcium carbonate scaling of seawater reverse osmosis systems, Desalination 284 (2012) 279–287. [22] X.M. Bao, M. Elimelech, Gypsum scaling and cleaning in forward osmosis: measurements and mechanisms, Environ. Sci. Technol. (2010) 2022–2028. [23] S. Zhao, L. Zou, Effects of working temperature on separation performance, membrane scaling and cleaning in forward osmosis desalination, Desalination 278 (2011) 157–164. [24] J. Zhang, W.L.C. Loong, S. Chou, C. Tang, R. Wang, A.G. Fane, Membrane biofouling and scaling in forward osmosis membrane bioreactor, J. Membr. Sci. 403-404 (2012) 8–14. [25] E. Arkhangelsky, F. Wicaksana, S. Chou, A.A. Al-Rabiah, S.M. Al-Zahrani, R. Wang, Effects of scaling and cleaning on the performance of forward osmosis hollow fiber membranes, J. Membr. Sci. 415-416 (2012) 101–108. [26] M. Zhang, J. Shan, C.Y. Tang, Gypsum scaling during forward osmosis process—a direct microscopic observation study, Desalin. Water Treat. 57 (2016) 3317–3327. [27] M.M. Zhang, D.X. Hou, Q.H. She, C.Y. Tang, Gypsum scaling in pressure retarded osmosis: experiments, mechanisms and implications, Water Res. 48 (2014) 387–395. [28] A. Achilli, T.Y. Cath, A.E. Childress, Selection of inorganic-based draw solutions for forward osmosis applications, J. Membr. Sci. 364 (2010) 233–241.

[29] L. Chekli, S. Phuntsho, H.K. Shon, S. Vigneswaran, J. Kandasamy, A. Chanan, A review of draw solutes in forward osmosis process and their use in modern applications, Desalin. Water Treat. 43 (2012) 167–184. [30] Q. Ge, M. Ling, T.-S. Chung, Draw solutions for forward osmosis processes: developments, challenges, and prospects for the future, J. Membr. Sci. 442 (2013) 225–237. [31] C. Klaysom, T.Y. Cath, T. Depuydt, I.F. Vankelecom, Forward and pressure retarded osmosis: potential solutions for global challenges in energy and water supply, Chem. Soc. Rev. 42 (2013) 6959–6989. [32] S. Zou, Y. Gu, D. Xiao, C.Y. Tang, The role of physical and chemical parameters on forward osmosis membrane fouling during algae separation, J. Membr. Sci. 366 (2011) 356–362. [33] Q.H. She, X. Jin, Q. Li, C.Y. Tang, Relating reverse and forward solute diffusion to membrane fouling in osmotically driven membrane processes, Water Res. 46 (2012) 2478–2486. [34] Y. Gu, L. Chen, J.W. Ng, C. Lee, V.W.C. Chang, C.Y. Tang, Development of anaerobic osmotic membrane bioreactor for low-strength wastewater treatment at mesophilic condition, J. Membr. Sci. 490 (2015) 197–208. [35] J.J. Qin, M.N. Wai, M.H. Oo, K.A. Kekre, H. Seah, Impact of anti-scalant on fouling of reverse osmosis membranes in reclamation of secondary effluent, Water Sci. Technol. (2016) 2767–2774. [36] J. Ren, J.R. McCutcheon, A new commercial thin film composite membrane for forward osmosis, Desalination 343 (2014) 187–193. [37] T.P.N. Nguyen, B.-M. Jun, J.H. Lee, Y.-N. Kwon, Comparison of integrally asymmetric and thin film composite structures for a desirable fashion of forward osmosis membranes, J. Membr. Sci. (2015). [38] J. Wei, C. Qiu, C.Y. Tang, R. Wang, A.G. Fane, Synthesis and characterization of flatsheet thin film composite forward osmosis membranes, J. Membr. Sci. 372 (2011) 292–302. [39] C.Y. Tang, T.H. Chong, A.G. Fane, Colloidal interactions and fouling of NF and RO membranes: a review, Adv. Colloid Interf. Sci. 164 (2011) 126–143. [40] C.Y. Tang, Q. She, W.C.L. Lay, R. Wang, A.G. Fane, Coupled effects of internal concentration polarization and fouling on flux behavior of forward osmosis membranes during humic acid filtration, J. Membr. Sci. 354 (2010) 123–133. [41] Q.H. She, Y.K.W. Wong, S.F. Zhao, C.Y. Tang, Organic fouling in pressure retarded osmosis: experiments, mechanisms and implications, J. Membr. Sci. 428 (2013) 181–189. [42] Q.H. She, X. Jin, C.Y. Tang, Osmotic power production from salinity gradient resource by pressure retarded osmosis: effects of operating conditions and reverse solute diffusion, J. Membr. Sci. 401–402 (2012) 262–273. [43] D.Z. Xiao, W. Li, S. Chou, R. Wang, C.Y. Tang, A modeling investigation on optimizing the design of forward osmosis hollow fiber modules, J. Membr. Sci. 392-393 (2012) 76–87. [44] S.K. Hong, M. Elimelech, Chemical and physical aspects of natural organic matter (NOM) fouling of nanofiltration membranes, J. Membr. Sci. 132 (1997) 159–181. [45] S.H. Yong, C.H. Lee, K.J. Kim, A.G. Fane, Effect of calcium ion on the fouling of nanofilter by humic acid in drinking water production, Water Res. 32 (1998) 2180–2186. [46] J.K. Fink, Chapter 11 – Scale Inhibitors, Hydraulic Fracturing Chemicals and Fluids Technology 2013, pp. 129–145 (Oxford). [47] K. Lutchmiah, J.W. Post, L.C. Rietveld, E.R. Cornelissen, EDTA: a synthetic draw solute for forward osmosis, Water Sci. Technol. 70 (2014) 1677–1682. [48] Q. Saren, C.Q. Qiu, C.Y. Tang, Synthesis and characterization of novel forward osmosis membranes based on layer-by-layer assembly, Environ. Sci. Technol. 45 (2011) 5201–5208. [49] C. Qiu, S. Qi, C.Y. Tang, Synthesis of high flux forward osmosis membranes by chemically crosslinked layer-by-layer polyelectrolytes, J. Membr. Sci. 381 (2011) 74–80. [50] L. Setiawan, R. Wang, K. Li, A.G. Fane, Fabrication of novel poly(amide-imide) forward osmosis hollow fiber membranes with a positively charged nanofiltrationlike selective layer, J. Membr. Sci. 369 (2011) 196–205. [51] Q. Yang, K.Y. Wang, T.S. Chung, Dual-layer hollow fibers with enhanced flux as novel forward osmosis membranes for water production, Environ. Sci. Technol. 43 (2009) 2800–2805. [52] J. Wei, C. Qiu, Y.-N. Wang, R. Wang, C.Y. Tang, Comparison of NF-like and RO-like thin film composite osmotically-driven membranes—implications for membrane selection and process optimization, J. Membr. Sci. 427 (2013) 460–471. [53] S. Qi, Y. Li, R. Wang, C.Y. Tang, Towards improved separation performance using porous FO membranes: the critical roles of membrane separation properties and draw solution, J. Membr. Sci. 498 (2016) 67–74. [54] S. Qi, Y. Li, Y. Zhao, W. Li, C.Y. Tang, Highly efficient forward osmosis based on porous membranes-applications and implications, Environ. Sci. Technol. 49 (2015) 4690–4695. [55] B.X. Mi, M. Elimelech, Gypsum scaling and cleaning in forward osmosis: measurements and mechanisms, Environ. Sci. Technol. 44 (2010) 2022–2028. [56] Y.S. Gu, Y.N. Wang, J. Wei, C.Y. Tang, Organic fouling of thin-film composite polyamide and cellulose triacetate forward osmosis membranes by oppositely charged macromolecules, Water Res. 47 (2013) 1867–1874. [57] C.Y. Tang, Y.N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes: II. Membrane physiochemical properties and their dependence on polyamide and coating layers, Desalination 242 (2009) 168–182. [58] A.G. Fane, C.Y. Tang, R. Wang, Membrane technology for water: microfiltration, ultrafiltration, nanofiltration, and reverse osmosis, in: P. Wilderer (Ed.), Treatise on Water Science, Academic Press, Oxford 2011, pp. 301–335.

Please cite this article as: M. Zhang, et al., Effect of reverse solute diffusion on scaling in forward osmosis: A new control strategy by tailoring draw solution chemistry, Desalination (2016), http://dx.doi.org/10.1016/j.desal.2016.08.014